Professional Documents
Culture Documents
Asian Development Bank Institute: ADBI Working Paper Series
Asian Development Bank Institute: ADBI Working Paper Series
IMPACTS OF COVID-19 ON
HOUSEHOLDS IN ASEAN COUNTRIES
AND THEIR IMPLICATIONS FOR
HUMAN CAPITAL DEVELOPMENT
No. 1226
March 2021
The Working Paper series is a continuation of the formerly named Discussion Paper series;
the numbering of the papers continued without interruption or change. ADBI’s working
papers reflect initial ideas on a topic and are posted online for discussion. Some working
papers may develop into other forms of publication.
Suggested citation:
Tel: +81-3-3593-5500
Fax: +81-3-3593-5571
URL: www.adbi.org
E-mail: info@adbi.org
Abstract
The outbreak of the COVID-19 virus and the resulting falls in demand due both to
uncertainty and policy interventions such as lockdowns, “social distancing,” and travel
restrictions are having a severe impact on Asian economies and hence on Asian households.
These negative impacts come through a variety of channels, including loss of employment or
reduced working hours, loss of sales and income of a household business, inability to travel
to work, increased need to stay at home to look after children or sick household members,
higher prices and/or lack of availability of staple items, reduced access to schooling, etc.
In order to better understand these impacts, we carried out computer-assisted telephone
interviews of households in eight ADB developing member countries: Cambodia, the Lao
People’s Democratic Republic, Indonesia, Malaysia, Myanmar, the Philippines, Thailand,
and Viet Nam. Our empirical results suggest that various household characteristics,
including household income class (before COVID-19), household demographic factors, and
COVID-19-induced factors such as having at least one person who lost their job or being
located in lockdown areas, all affected the likelihood of a decline in income. In all countries,
having at least one person who lost their job or had reduced working time increases
the likelihood of experiencing financial difficulties by 17 percentage points. About 27% of
children who stopped attending school could not fully participate in online learning programs
due to weak/insufficient internet connections and a lack of digital devices. Two COVID-19-
related factors—having at least one person who lost their job or had working hours reduced
and experiencing financial difficulties—significantly affect the intensity of online classes
taken by children in an average household.
2
ADBI Working Paper 1226 Morgan and Trinh
Contents
1. INTRODUCTION............................................................................................................1
REFERENCES ........................................................................................................................30
3
ADBI Working Paper 1226 Morgan and Trinh
1. INTRODUCTION
The outbreak of the COVID-19 virus and the resulting falls in demand due both to
uncertainty and policy interventions such as lockdowns, “social distancing,” and travel
restrictions are having a severe impact on Asian economies and hence on Asian
households. These negative impacts come through a variety of channels, including loss
of employment or reduced working hours, loss of sales and income of a household
business, inability to travel to work, increased need to stay at home to look after
children or sick household members, higher prices and/or lack of availability of staple
items, reduced access to schooling, etc. Low-income and vulnerable groups are likely
to be particularly susceptible due to a lack of resources they can draw on during
periods of reduced income. In order to develop appropriate policy responses, it is
necessary to obtain a correct understanding of the current situation of households. As
part of the Asian Development Bank’s overall strategy to deal with the current crisis,
ADBI has carried out surveys of households to better understand the size, aspects, and
incidence of impacts on them, especially vulnerable groups. In particular, this study
focuses on the implications for human capital formation of reduced access to education
due to reduced school attendance because of the pandemic and inability to access
online learning due to insufficient internet connectivity and a lack of devices to connect.
Specifically, we carried out computer-assisted telephone interviews (CATIs) and online
surveys of households in eight ADB developing member countries: Cambodia, the Lao
People’s Democratic Republic (Lao PDR), Indonesia, Malaysia, Myanmar, the
Philippines, Thailand, and Viet Nam. (Face-to-face surveys were impractical due to the
lockdowns being implemented in response to the pandemic.) The surveys were carried
out from the end of May to the end of July. Representative samples of 1,000
households in each country were surveyed.
2. LITERATURE REVIEW
Since the first case was reported in Hubei province in the People’s Republic of China
(PRC) in December 2019, the new coronavirus has spread rapidly throughout the
world, and the World Health Organization (WHO) declared it a global pandemic in
March 2020 (Cucinotta and Vanelli 2020). The disease has since become one of the
largest public health crises in history, and it has led to economic and social crises
as well. In some ASEAN economies, the first cases of COVID-19 were reported quite
early, in January 2020. Since then, the disease has continued to spread in the region.
ASEAN economies’ governments have instituted various restrictions to contain the
spread of the disease within their boundaries, such as movement restrictions, border
closures, social distancing, quarantine, and closure of nonessential services.
Implementation of such stringent measures could disrupt the economy and ultimately
people’s living standards. Reports show that the region will face disruptions in several
ways, including the loss of jobs and income, and potentially reduced accumulation
of human capital for children (Demeke and Kariuki 2020; ILO 2020; UN-Habitat and
WFP 2020).
Many studies have shown profound impacts of the COVID-19 pandemic on economic
activities. McKibbin and Fernando (2020) argue that pandemics can affect households,
governments, and businesses through increased business costs, increased public
healthcare expenditures, changes in labor supply due to mortality and morbidity, etc.
COVID-19-related restrictions have obstructed all stages of the food supply chain,
including production, distribution, processing, and consumption (Siche 2020; Torero
1
ADBI Working Paper 1226 Morgan and Trinh
2020), and have caused damage to perishable agricultural commodities such as meat
and vegetables (Nicola et al. 2020).
While various studies have examined potential impacts of the pandemic on global and
national economic indicators such as poverty, government expenditures, GDP growth,
budget deficits, and employment (ILO 2020; Nicola et al. 2020; Sumner, Hoy, and
Ortiz-Juarez 2020; UN-Habitat and WFP 2020), there is limited information on how the
pandemic and associated lockdown policies are affecting individuals at the household
level. Economic effects of a pandemic may disproportionately impact members of
society, depending on their income status, livelihood strategies, access to markets, etc.
Therefore it is important to understand the household-level impacts as well as the
support mechanisms that could contribute to income smoothing.
Kansiime et al. (2020) assessed implications of the COVID-19 pandemic for household
income and food security in two East African countries, using online survey data from
442 respondents. They found that more than two-thirds of the respondents experienced
income shocks due to the COVID-19 crisis. Food security and dietary quality worsened:
the proportion of food-insecure1 respondents increased by 38% and 44% in Kenya and
Uganda, respectively. Results from probit regressions show that the low-income
households and those dependent on labor income were more vulnerable to income
shocks and consumed less food during the COVID-19 pandemic compared to other
respondent categories.
Ceballos, Kannan, and Kramer (2020) and Harris et al. (2020) studied the impacts of
COVID-19 and related restrictions on smallholders in India and reported a large degree
of heterogeneity in the impacts of COVID-19 responses on agricultural activity, income,
and food security. Ceballos, Kannan, and Kramer (2020) analyzed data from phone-
based surveys on disruptions to agricultural production and food security administered
to 1,515 smallholder producers in the states of Haryana and Odisha, India. They found
substantial heterogeneity in how the lockdown affected farmers in these two states,
which was likely related to existing structural differences in market infrastructure and to
differences in state-specific COVID-19-related policies. Harris et al. (2020) investigated
effects of the COVID-19-induced shock on the production, sales, prices, incomes, and
diets of vegetable farmers in India as both producers and consumers of nutrient-dense
foods. They undertook a rapid telephone survey with 448 farmers in four states. They
found that a majority of farmers reported negative impacts on production, sales, prices,
and incomes. Over 80% of farmers reported some decline in sales, and over 20%
reported very large declines (sold almost nothing). Price reductions were reported by
over 80% of farmers, and reductions of more than half by 50% of them. Similarly, farm
income reportedly dropped for 90% of farmers, and by more than half for 60% of them.
The data also showed a higher level of vulnerability among female farmers in terms of
both livelihoods and diet, and differential effects on smaller and larger farms, which
suggests that different farms may require different types of support in order to continue
to operate. Farms reported diverse coping strategies to maintain sales, though often
with negative implications for reported incomes. The ability to consume one’s own
produce may help to support diets.
We add to the growing body of literature on the impacts of the COVID-19 pandemic by
examining its implications for household income, household financial viability, and
schooling decisions, covering eight Southeast Asian countries: Cambodia, the Lao
PDR, Indonesia, Malaysia, Myanmar, the Philippines, Thailand, and Viet Nam. The
1 Food insecurity is defined as the disruption of food intake or eating patterns because of a lack of money
and other resources.
2
ADBI Working Paper 1226 Morgan and Trinh
3
ADBI Working Paper 1226 Morgan and Trinh
4
ADBI Working Paper 1226 Morgan and Trinh
5
ADBI Working Paper 1226 Morgan and Trinh
Source: Authors’ calculation based on ADBI’s database. Relative to base period of end-2019.
2 The poorest group comprises households in income classes 3, 4, and 5 in Cambodia, Indonesia, and
Thailand; households in income classes 4, 5, and 6 in the Lao PDR and Viet Nam; households in
income classes 3 and 4 in Malaysia and Myanmar; and households in income classes 5 and 6 in
Philippines.
6
ADBI Working Paper 1226 Morgan and Trinh
Notes: Poorer are households in Socio-economic Classes 3, 4, 5 and 6 while richer are households in the Socio-
economic Classes 1 and 2.
Source: Authors’ calculation based on ADBI’s database.
Figure 5 further documents sources of income decline for each country in our sample.
Household business/self-employed income fell the most in all countries except
Indonesia, where agricultural/fishery income showed the highest percentage of decline.
7
ADBI Working Paper 1226 Morgan and Trinh
𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑒𝑒𝑖𝑖 = 𝛼𝛼0 + 𝛼𝛼1 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 + 𝛼𝛼2 𝐻𝐻𝐻𝐻𝑖𝑖 + 𝛼𝛼3 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖 + 𝜖𝜖𝑖𝑖 (1)
3 Due to differences across countries in categorizing income classes (some countries have four classes,
some have six classes, and some use wealth rather than income to construct income classes),
we reconstruct our own income classes. There are four groups of income by the income quartile
(25%, 50%, and 75%).
8
ADBI Working Paper 1226 Morgan and Trinh
that the income inequality may be widened after the COVID-19 pandemic in the
Philippines and narrowed in Cambodia.
Different sources of income may also have different effects on the likelihood of a
decline in income. In general, those households with income from agricultural
production and from household businesses or self-employment are more likely to
experience a decline in income. This situation is observed in nearly all countries in
our sample. Cambodian, Indonesian, and Filipino households that have income from
agricultural production are not different from other households that do not have this
source of income. This may suggest that people in these ASEAN economies lack a
well-developed social safety net to support them when there is a shock. Households
whose income is mainly from salaries are less likely to experience income declines.
However, this relationship is only observed in three countries: Indonesia, the Lao PDR,
and Thailand. In other countries, the likelihood of experiencing a decline in income is
not affected by whether the households have wages/salaries from formal sectors as an
income source. This finding shows that households with remittance income are not
different from other households that do not have remittance income. Also, households
with income from other sources, such as house rental, moneylending, or receiving
regular government support/aid, are less likely to suffer from income declines, but this
is only observed in three countries: Myanmar, Indonesia, and the Philippines.
The education level of the household head in general is negatively associated with the
likelihood of experiencing an income decline. For example, household heads with a
high school degree have a lower probability of income decline by 2.5 percentage points
than those who have lower degrees. The figure for those who have a higher education
level than high school is 8.7 percentage points lower. However, the role of the
household head’s education level is not observed in all countries in our sample. In the
Lao PDR, Myanmar, the Philippines, and Viet Nam, the household head’s education
did not determine whether the household experienced a decline in income.
Household head’s age is not associated with the COVID-19-induced income decline in
our pooled sample, but it has some effects in some countries. In Cambodia and
Malaysia, the likelihood of experiencing a decline in income is lower for households
whose head is older than 50, but in the Lao PDR and the Philippines, households
whose head is aged between 30 and 39 may be more likely to experience a decline
in income than younger or older households. The effects of other household
characteristics such as household size, dependency ratio, and house location are
generally not found to be significant in most countries.
Two variables that are directly related to the COVID-19 pandemic are significantly
associated with income declines. Being located in a lockdown area increases the
likelihood of having an income decline in half of the countries (Indonesia, the Lao PDR,
the Philippines, and Viet Nam). While the effects of this in Indonesia and the Lao PDR
are quite small and marginally significant, the effect is much stronger in other two
countries, especially in the Philippines. Malaysia is an interesting case, where those
living in lockdown areas have a lower likelihood of experiencing income decline. A
potential explanation for this negative relation is that those living in lockdown areas
receive government aid, which may help to keep their income stable.
Having someone in the household who has lost their job (or has had to reduce working
hours) is positively associated with income decline in all countries. However, the effect
differs widely, with the smallest effect being observed in Indonesia and the largest
in Thailand.
9
ADBI Working Paper 1226 Morgan and Trinh
10
ADBI Working Paper 1226 Morgan and Trinh
Table 4 continued
(6) (7) (8) (9)
Myanmar Philippines Thailand Viet Nam
SEC 2 0.072** 0.134** –0.018 0.049
[0.036] [0.064] [0.048] [0.051]
SEC 3 0.039 0.152** 0.065* 0.044
[0.036] [0.063] [0.039] [0.048]
SEC 4 0.016 0.212*** 0.065 0.042
[0.040] [0.066] [0.045] [0.046]
Income source: Agriculture related 0.066* 0.043 0.147*** 0.086**
[0.036] [0.035] [0.034] [0.039]
Income source: Household business/ 0.107*** 0.089*** 0.258*** 0.192***
self-employed [0.032] [0.030] [0.033] [0.034]
Income source: Wage –0.037 0.029 –0.094** –0.027
[0.030] [0.032] [0.037] [0.036]
Income source: Remittance –0.026 0.055 0.056 –0.184
[0.070] [0.040] [0.063] [0.123]
Income source: Others –0.117** –0.151*** –0.036 –0.072
[0.047] [0.040] [0.047] [0.047]
HH head with high school degree 0.002 –0.030 –0.040 0.015
[0.031] [0.028] [0.031] [0.037]
HH head with more than high school degree –0.038 0.009 –0.057** –0.041
[0.035] [0.031] [0.028] [0.041]
Household head being male 0.008 –0.013 0.004 –0.040
[0.036] [0.026] [0.024] [0.031]
Household head aged 30‒39 –0.034 0.086** –0.009 –0.014
[0.044] [0.039] [0.047] [0.054]
Household head aged 40‒49 –0.047 0.062 –0.021 –0.043
[0.045] [0.041] [0.044] [0.055]
Household head aged 50‒59 –0.012 0.063 –0.001 –0.098*
[0.042] [0.041] [0.044] [0.059]
Household head aged 60 or over –0.050 0.067 –0.036 –0.070
[0.049] [0.053] [0.050] [0.063]
Household size –0.006 0.003 –0.001 0.008
[0.007] [0.005] [0.007] [0.011]
Dependency ratio 0.183** –0.042 0.013 0.045
[0.072] [0.060] [0.049] [0.076]
Located in rural areas –0.016 –0.035 –0.004 –0.045
[0.028] [0.024] [0.025] [0.032]
Located in lockdown areas 0.036 0.992*** –0.008 0.178***
[0.059] [0.043] [0.029] [0.066]
At least one HH member lost job or reduced 0.187*** 0.188*** 0.465*** 0.278***
work [0.027] [0.031] [0.026] [0.030]
Intercept 0.624*** –0.504*** 0.475*** 0.294***
[0.079] [0.094] [0.077] [0.108]
Number of observations 999 998 988 914
Notes: Standard errors are in brackets. *, **, and *** denote statistically significant at 10%, 5%, and 1% confidence
levels, respectively. We control for country dummies in the pooled sample (column 1).
Source: Authors’ estimation using ADBI database.
11
ADBI Working Paper 1226 Morgan and Trinh
Only 37% of households in our sample reported that their expenditure declined, which
was much lower than the approximately 73% of households that experienced a decline
in incomes (Figure 6). Nearly 30% of households actually increased their expenditure.
There are several reasons for the expenditure increases. Figure 7 reports unexpected
expenditures by households during the COVID-19 period. Nearly 80% and 50%
of households had to spend more for healthcare products and household cleaning
products, respectively. Unexpected (higher) utility bills were also a source of increasing
expenditure. Nearly 34.4% of households had to pay more for food.
12
ADBI Working Paper 1226 Morgan and Trinh
𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑓𝑓𝑖𝑖 = 𝛽𝛽0 + 𝛽𝛽1 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 + 𝛽𝛽2 𝐻𝐻𝐻𝐻𝑖𝑖 + 𝛽𝛽3 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖 + 𝜂𝜂𝑖𝑖 (2)
13
ADBI Working Paper 1226 Morgan and Trinh
Table 5 presents our estimation results. The first column reports the results using
pooled data and the subsequent columns are the results for the individual countries.
We found that, in contrast to the results presented in Table 1, the income class of
household on average is associated with the likelihood of having financial difficulty.
Households in the lowest income class are more likely to get into financial difficulty
than households in the highest income class by about 20 percentage points. The
differences in the likelihood among the richest group and the poorest group ranged
from 14.4 percentage points in Viet Nam to about 40 percentage points in the
Philippines. This may be due to the fact that some households suffered from income
decline, but they also had to increase their expenditure. The likelihood of getting into
financial difficulty is also higher for other income classes in our pooled sample and
in some countries, except for Cambodia and the Lao PDR, where we only observed
differences between the richest and the poorest income groups.
Different sources of income may also have different effects on the likelihood of getting
into financial difficulty. In general, households with income from agricultural production,
household businesses, or self-employment and remittances are not different from
households that do not have such income sources. However, households with income
from wages and salaries and with other income sources are less likely to have financial
difficulty than those households that do not have such income. This finding, to some
extent, is corroborated by the results presented in Table 4. We also find that the effect
of income source differs by country. For example, Indonesian households with income
from agricultural production, household business/self-employment, and wages/salaries
are more likely to be in financial difficulty while this is not the case for Malaysian
households.
The education level of the household head in general is negatively associated with the
likelihood of getting into financial difficulty. For example, household heads with a high
school degree have a lower probability of having financial difficulty by 2.7 percentage
points than those who have a lower level of education. The respective figure for those
who have a higher education level than high school is 7.8 percentage points. However,
the effect of the household head’s education level is only significant in Indonesia,
Malaysia, and Viet Nam. Households with a household head aged over 50 are less
likely to be in financial difficulty than other households in the pooled sample, and those
in Cambodia, Malaysia, Myanmar, and Viet Nam. In contrast to our results regarding
the likelihood of having an income decline, household size and household dependency
ratio are positively associated with the likelihood of getting into financial difficulty. For
example, having one more person in the household increases the likelihood of being
in financial difficulty by 1 percentage point (2 percentage points in Indonesia and
Thailand). The dependency ratio is also positively associated with a higher likelihood of
getting into financial difficulty in five out of eight countries.
The estimation results also suggest that having someone in the household who has
lost their job or had to reduce working hours is positively associated with the likelihood
of being in financial difficulty in all countries, although the size of the effect varies.
Being located in a lockdown area also increases the likelihood of having financial
difficulty in Cambodia, the Lao PDR, and Myanmar as expected, but the effect is
negative in Viet Nam.
14
ADBI Working Paper 1226 Morgan and Trinh
15
ADBI Working Paper 1226 Morgan and Trinh
Table 5 continued
(6) (7) (8) (9)
Myanmar Philippines Thailand Viet Nam
SEC 2 0.012 0.211*** –0.022 0.101*
[0.033] [0.073] [0.057] [0.052]
SEC 3 0.133*** 0.339*** 0.133*** 0.092*
[0.037] [0.069] [0.046] [0.050]
SEC 4 0.222*** 0.398*** 0.204*** 0.144***
[0.043] [0.070] [0.052] [0.052]
Income source: Agriculture related –0.096*** –0.062* 0.089** 0.005
[0.034] [0.036] [0.037] [0.041]
Income source: Household business –0.024 –0.025 0.142*** 0.046
[0.033] [0.034] [0.035] [0.036]
Income source: Wage 0.032 –0.083** –0.052 –0.050
[0.031] [0.035] [0.036] [0.039]
Income source: Remittance –0.130*** –0.046 0.065 0.057
[0.046] [0.044] [0.070] [0.144]
Income source: Others 0.022 –0.039 –0.097* –0.145***
[0.042] [0.033] [0.056] [0.046]
HH head with high school degree –0.012 –0.016 –0.037 –0.034
[0.034] [0.024] [0.035] [0.040]
HH head with more than high school –0.041 –0.016 –0.042 –0.161***
degree [0.036] [0.028] [0.033] [0.042]
Household head being male –0.083** –0.022 –0.002 –0.146***
[0.041] [0.024] [0.028] [0.033]
Household head aged 30‒39 –0.031 0.012 –0.018 0.037
[0.051] [0.036] [0.052] [0.058]
Household head aged 40‒49 –0.011 0.009 –0.001 –0.063
[0.051] [0.037] [0.050] [0.059]
Household head aged 50‒59 –0.083* 0.019 –0.041 –0.108*
[0.049] [0.040] [0.050] [0.061]
Household head aged 60 or over –0.125** –0.093* –0.103* –0.151**
[0.053] [0.051] [0.060] [0.067]
Household size 0.011 0.001 0.018** 0.002
[0.008] [0.005] [0.008] [0.010]
Dependency ratio 0.146* 0.134** 0.034 0.079
[0.076] [0.054] [0.056] [0.079]
Located in rural areas 0.065** –0.039* 0.108*** –0.097***
[0.031] [0.023] [0.029] [0.035]
Located in lockdown areas 0.287*** –0.020 0.040 –0.104*
[0.083] [0.038] [0.030] [0.061]
At least one HH member lost job or 0.077*** 0.143*** 0.243*** 0.172***
reduced work [0.029] [0.029] [0.028] [0.032]
Intercept 0.173** 0.519*** 0.393*** 0.560***
[0.081] [0.093] [0.087] [0.110]
Number of observations 997 998 988 914
Notes: Standard errors are in brackets. *, **, and *** denote statistically significant at 10%, 5%, and 1%, respectively.
We control for country dummies in the pooled sample.
Source: Authors’ estimation using ADBI database.
16
ADBI Working Paper 1226 Morgan and Trinh
Nearly 80% of respondents who suffered from financial difficulty reduced their food
consumption. This may have some implications for food security. Other items that
people reduced include utilities and nonessential daily expenditures (Figure 10).
17
ADBI Working Paper 1226 Morgan and Trinh
Figure 11 shows the time period for which households have enough resources to cover
daily necessities if they lose all of their income sources. On average, nearly 50% of
households in our sample could only survive less than a month. About 86.6% of
Indonesian households reported that they could only afford about two weeks if they did
not have any income. The situation is also serious in the Philippines, with less than
30% having enough resources to cover necessary expenditures for more than a month.
This suggests that if the COVID-19 pandemic is prolonged, many households in these
countries may suffer from hunger and potentially increased poverty.
Figure 11: Duration over which Resources Could Cover Basic Necessities,
% of Households
𝑉𝑉𝑉𝑉𝑎𝑎𝑖𝑖 = 𝛾𝛾0 + 𝛾𝛾1 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 + 𝛾𝛾2 𝐻𝐻𝐻𝐻𝑖𝑖 + 𝛾𝛾3 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖 + 𝜉𝜉𝑖𝑖 (3)
in which 𝑉𝑉𝑉𝑉𝑎𝑎𝑖𝑖 is the duration of financial viability of household i; 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 , 𝐻𝐻𝐻𝐻𝑖𝑖 , and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖
are as in Equation (1), and 𝜉𝜉𝑖𝑖 is the error term. We also controlled country dummies in
our estimation using pooled data. Based on Likert-type measures, we constructed
three dummy variables indicating the duration of financial viability: (1) being able to
survive up to a month; (2) being able to survive from one month up to three months;
and (iii) being able to survive more than three months. Due to the nature of our
dependent variable (Likert-type variables), the ordered probit estimation method
is used.
Table 6 presents our estimation results using pooled sample data. The marginal effects
are calculated for each of three possible outcomes. We find that household income
class has a statistically significant correlation with the duration of financial viability.
Households in lower income classes have a higher likelihood of being financially viable
for less than one month than those in the highest income class by 10, 15, and 21
percentage points, respectively. Similarly, those in the highest income class have a
18
ADBI Working Paper 1226 Morgan and Trinh
higher chance of being financially viable for more than three months by about 10, 13,
and 17 percentage points compared with the low-income groups. This result suggests
that if the COVID-19 pandemic is prolonged, poorer households will suffer much more
than richer households.
19
ADBI Working Paper 1226 Morgan and Trinh
The source of income also matters for the duration of financial viability. Households
with income from agricultural production, household business, and other sources
tend to have a longer duration of financial viability. Therefore, despite the fact that
households with income from agricultural production and from household business/self-
employment are more likely to experience a decline in income (as shown in Table 1),
these households have a higher probability of financial viability than other households.
Household characteristics also matter for the duration of financial viability. Households
whose head has an education level of at least a high school degree are more likely to
be financially viable for a longer time than those whose education level is lower. For
example, the former group is more likely to be financially viable for more than three
months than the latter by 3.5 and 7.3 percentage points. Households headed by a man
are also more (less) likely to be able to survive financially for more than three months
(less than a month) than households headed by a woman. Household size and the
dependency ratio are also associated with the duration of financial viability.
With regard to COVID-19-induced variables, both living in a lockdown area and having
at least one person who lost their job or had working time reduced are associated with
the duration of financial viability. Households in the latter group are more likely to
survive less than a month and less likely to survive from one to three months and for
more than three months by 4.9, 1.3, and 3.6 percentage points, respectively, than other
households.
20
ADBI Working Paper 1226 Morgan and Trinh
On average, 44.4% of employees in our samples had either lost their jobs (temporarily
or permanently) or had experienced a workload cut. Notably, 73.5% of Filipino
employees in our samples had either lost their jobs or had to reduce their working time.
In Viet Nam, Thailand, Myanmar, and Malaysia, the figures were high, in the range of
45%‒50%, while the share in Indonesia was relatively low.
We further looked at types of job loss or workload reduction. Figure 14 shows types
of job loss. It should be noted that, in contrast to Figure 13, which looks at total
employment (by summing up the total number of people who lost jobs or had reduced
working time), Figure 14 shows types of jobs lost at the household level. This is
because we cannot distinguish between types of job lost and working time reduced
among household members (i.e., we cannot identify how many people in a certain
household lost jobs and how many of them are forced to reduce their working time).
Among all households, 8.8% had at least one person who lost their job permanently,
10.1% had at least one person who lost their job temporarily, 6.6% had at least one
person who was forced to reduce their working hours, and 8% had at least one person
who was forced to reduce their working days.
The pattern of job losses is quite similar in Indonesia, Malaysia, Myanmar, Thailand,
and Viet Nam. The proportion who lost jobs permanently accounted for about a quarter
of all households that have at least one person who lost their job or were forced to
reduce working time (i.e., accounting for about one tenth of the total sample). The
proportion of households with at least one person who lost their job or had work time
reduced is lowest in Cambodia and the Lao PDR. More than 70% of households in the
Philippines had at least one person who lost their job or had to reduce working time.
More strikingly, the proportion of those in the Philippines who lost their job permanently
or temporarily is very high (22.5% and 28.5%, respectively). While the proportion of
households with at least one person who lost their job in the Lao PDR is low, the
proportion of those who lost their jobs permanently is quite high (12.5%), second only
to the Philippines.
21
ADBI Working Paper 1226 Morgan and Trinh
Table 8 shows the breakdown in school attendance by country and type of household.
There are no significant differences in school attendance between children in richer
and poorer households or between rural and urban households. Even in Thailand, the
children in rural areas were more likely to attend school than children in urban areas.
This may be due to the fact that most of the COVID-19 cases identified in Thailand
were in urban areas.
22
ADBI Working Paper 1226 Morgan and Trinh
During the pandemic, many schools closed physically and offered online classes
(Figure 15). For example, 94% of Indonesian children studied in schools that offered
online classes. In other countries, including Thailand, Viet Nam, Malaysia, and
Cambodia, around 70%‒77% of children studied in schools with online classes. The
figure is slightly lower in the Philippines at only 57% of children. Conversely, most
children in the Lao PDR and Myanmar who were still going to school still went to school
rather than taking online classes. The figures presented in Table 8 and Figure 15
indicate that all Lao PDR pupils are still going to school and only 10% of children are
studying in online classes. This suggests that most Lao PDR pupils continued going to
school during the pandemic. Meanwhile, only 48% of Myanmarese pupils continued
their education during the pandemic. This may be because only a small number of
schools in Myanmar can offer online courses. Our data indicate that only 6% of children
studied in schools with online classes there.
Even when schools offered online classes, not all children attended them. About 8%
did not attend any online classes, 19% attended only a few, and 16% attended some
but not all (Figure 16). A high proportion of children in Indonesia and Viet Nam
attended all online classes if offered (79%‒80%), while the figure was very low in the
Philippines (only 21%) and only moderate in Thailand, Myanmar, and Malaysia. Fully
23
ADBI Working Paper 1226 Morgan and Trinh
35% of Filipino children did not attend any online classes. In Myanmar, nearly 70% did
not fully attend online classes.
Table 9 shows the share of children from different household types attending all online
classes if offered. Children from richer households tended to attend all online class
more than those from poorer households in all countries except Cambodia and the
Lao PDR. It should be noted that the differences between richer and poorer
households are larger for the subgroup of those who did not attend (or attended only a
few) online classes (data available upon request). There were no significant differences
in attendance of online classes between rural and urban households except in the
|Lao PDR and Myanmar.
24
ADBI Working Paper 1226 Morgan and Trinh
𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑦𝑦𝑖𝑖 = 𝜃𝜃0 + 𝜃𝜃1 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 + 𝜃𝜃2 𝐻𝐻𝐻𝐻𝑖𝑖 + 𝜃𝜃3 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖 + 𝑒𝑒𝑖𝑖 (4)
in which intensity takes the value of either (1) not taking any online class; (2) taking a
few online classes; (3) taking some, but not all online classes; and (4) taking all online
classes; 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 ; 𝐻𝐻𝐻𝐻𝑖𝑖 , and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖 are as in Equation (1) and 𝑒𝑒𝑖𝑖 is the error term. Since
our dependent variable is a Likert-type variable, the ordered probit estimation method
is used. However, the results of this regression analysis should be interpreted
cautiously since the sample used is limited to those children who study in schools that
offer online classes.
Table 10 presents the results for the determinants of the intensity of online course
attendance. The marginal effects are calculated for each of the four possible outcomes.
Household income class has a statistically significant correlation with the intensity
of online course attendance. For example, those in lower-income classes are more
likely not to take any online classes and less likely to take all classes than those in the
highest income class (SEC 1 as the reference group). However, we did not see any
significant difference between households in the lowest SEC classes and the highest
SEC classes.
Household characteristics also matter for the intensity of online class taking. Children
from households whose head has an education level higher than high school education
were less likely not to take any online classes and more likely to take all online classes
than children in households with lower education levels (our reference group), although
this was only significant at the 10% level. We do not find any difference between
children living in households whose head has only a high school education and
children in the reference group. The education level of the household head is also not
significantly associated with whether children take few or some online classes. The
gender of the household head is also not correlated with the intensity of the online
class attendance of children. However, the age of the household head is related to
the level of online course attendance. Children in households whose heads are older
than 39 are less likely not to take any online course and more likely to take all online
courses offered. This may be because the older the household head is, the more
resilient the household is to external shocks and therefore children are given more
resources for their education.
With regard to COVID-19-induced variables, living in a lockdown area is positively
associated with taking all online courses, but the relationship is only significant at the
10% level. A change in income is not significantly correlated with the level of online
courses taken, but living in households with at least one person who has lost their job
or experienced a working time reduction is positively associated with a lower intensity
of online course attendance. A similar and statistically stronger pattern is observed for
children living in households experiencing financial difficulty.
25
ADBI Working Paper 1226 Morgan and Trinh
We examined the reasons why some students did not fully attend the online classes,
given that the school offered them (Figure 17). There are four major reasons: doing
household chores; lack of computer/iPad/tablet; lack of internet connection; and
weak/unstable internet connection. On average, about a third of households with
children who did not attend online classes fully did not have facilities for online learning
(i.e., devices and connections). This situation was most common in the Philippines
26
ADBI Working Paper 1226 Morgan and Trinh
where more than 70% of households reported that their children did not attend online
classes because they did not have a computer or tablet for their children to use, and
nearly 80% said that they did not have internet connection.
Figure 17: Reasons for Not Attending Online Classes Fully (if School Offered),
% of Households
We further examine underlying factors correlated with why some children could not fully
take the online courses. We estimate the following equation:
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑖𝑖 = 𝜕𝜕0 + 𝜕𝜕1 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 + 𝜕𝜕2 𝐻𝐻𝐻𝐻𝑖𝑖 + 𝜕𝜕3 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖 + 𝑢𝑢𝑖𝑖 (5)
in which 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑖𝑖 is the reason why household i’s children cannot take all online classes
(when offered); 𝑆𝑆𝑆𝑆𝐶𝐶𝑖𝑖 , 𝐻𝐻𝐻𝐻𝑖𝑖 , and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷𝑖𝑖 are as in Equation (1) and 𝑢𝑢𝑖𝑖 is the error
term. The dependent variable takes one of four alternatives: (i) doing household
chores/working with the family; (ii) not having a computer/iPad/tablet; (iii) not having
internet connection; and (iv) unstable internet connection.4
Table 11 presents the results for factors related to each of the four reasons why
children in an average household could not attend all online classes. Household
income class is correlated with the likelihood of not taking all classes because of doing
household chores, not having a computer/iPad/tablet, or not having an internet
connection, among others. Household income is not related to the reason regarding the
unstable/weak internet connection. We also find that household characteristics play a
role for some reasons why children did not take all classes. For example, those who
did not take all online classes due to not having a computer/iPad/tablet or not having
an internet connection tend to came from households whose head has a lower
education level. Children in households whose head is older were less likely not to
attend online classes because of doing household chores than children in households
with a younger household head. An unstable internet connection was less likely to be
the reason for not attending all classes for children whose parents are aged 30‒49.
Children in households with a high dependency ratio were more likely not to attend
online classes due to not having a computer/iPad/tablet or not having an internet
4 The survey also asked for other reasons but, due to a lack of clear information about this type of reason,
we did not use other reasons as an alternative reason for not taking online classes.
27
ADBI Working Paper 1226 Morgan and Trinh
connection. Living in a rural area was more likely to be the reason for not attending all
classes due to not having an internet connection.
28
ADBI Working Paper 1226 Morgan and Trinh
difficulty were more likely not to attend all online classes due to not having a
computer/iPad/tablet and not having an internet connection than children in households
not experiencing financial difficulty.
8. CONCLUDING REMARKS
The outbreak of the COVID-19 virus and the resulting falls in demand due both to
uncertainty and policy interventions such as lockdowns, “social distancing,” and travel
restrictions are having a severe impact on Asian economies and hence on Asian
households. In this paper, we implemented household surveys in eight ASEAN
countries (Cambodia, Indonesia, the Lao PDR, Malaysia, Myanmar, the Philippines,
Thailand, and Viet Nam) to understand the effects of the COVID-19 pandemic. The
samples were representative of the income classes, and rural and urban population in
each country.
The paper presents several interesting results. First, most households in all countries
experienced significant declines in income and employment, although there was
significant variation by country. Nearly three quarters of households experienced a
declining income (mostly in the range of 26%‒50% compared with the base period).
Income from all sources declined, but that of household enterprise/self-employment fell
most. About a third had household members who lost their jobs or had to reduce their
workload. Our empirical results suggest that various household characteristics,
including household income class (before the COVID-19 pandemic) and household
demographic factors, affect the likelihood of a decline in income. COVID-19-induced
factors such as having at least one person who lost their job or being located in
lockdown areas also affect household income.
Second, nearly half of households experienced financial difficulties. Nearly all
households experiencing financial difficulties had to reduce consumption, about half
drew down cash and savings, while roughly a third did the following: (i) borrowed from
friends or relatives; (ii) delayed payments and debt repayment; and (iii) applied for
government aid. Furthermore, the share of households that could only survive without
any income for less than a month is quite high, at about 50% of all households in our
sample. In addition to household characteristics such as household head’s education,
age, and gender, COVID-19-induced factors also play an important role in the
likelihood of experiencing financial difficulties and the duration of financial viability. In all
countries, having at least one person who lost their job or had reduced working time
increases the likelihood of experiencing financial difficulties by 17 percentage points.
This factor is statistically significant in all eight countries.
Third, about 27% of children stopped going to school (especially in Myanmar and the
Philippines) as a result of the pandemic. Many schools offered online classes, but there
was great variation by country. About 27% of children who stopped attending school
could not fully participate in online learning programs due to weak/insufficient internet
connections and a lack of digital devices. Of all the factors possibly determining the
intensity of online class taking, two COVID-19-related factors—having at least one
person who lost their job or had working hours reduced and was experiencing financial
difficulties—significantly affect the intensity of online classes taken by children in an
average household.
29
ADBI Working Paper 1226 Morgan and Trinh
REFERENCES
Ceballos, F., Kannan, S. and Kramer, B., 2020. Impacts of a national lockdown on
smallholder farmers’ income and food security: Empirical evidence from two
states in India. World Development, 136, p.105069.
Cucinotta, D. and Vanelli, M., 2020. WHO declares COVID-19 a pandemic. Acta Bio
Medica: Atenei Parmensis, 91(1), p.157.
Demeke, M. and Kariuki, J.F., 2020. Assessing the impact of COVID-19 on food and
nutrition security and adequacy of responses in Kenya. Mimeo.
Hale, T., Petherick, A., Phillips, T. and Webster, S., 2020. Variation in government
responses to COVID-19. Blavatnik School of Government Working Paper, 31.
Available at https://ourworldindata.org/grapher/covid-stringency-index.
Harris, J., Depenbusch, L., Pal, A.A., Nair, R.M. and Ramasamy, S., 2020. Food
system disruption: Initial livelihood and dietary effects of COVID-19 on
vegetable producers in India. Food Security, 12(4), pp.841–851.
International Monetary Fund, 2020. Policy Responses to COVID-19. Available at
https://www.imf.org/en/Topics/imf-and-covid19/Policy-Responses-to-COVID-19.
Accessed Nov. 15, 2020.
International Labor Organization, 2020. COVID-19 and the world of work: Impact and
policy responses. ILO Monitor (1st Ed.), Geneva (2020).
Kansiime, M.K., Tambo, J.A., Mugambi, I., Bundi, M., Kara, A. and Owuor, C., 2020.
COVID-19 implications on household income and food security in Kenya and
Uganda: Findings from a rapid assessment. World Development, 137, p.105199.
McKibbin, W.J., and Fernando, R. (2020). The global macroeconomic impacts of
COVID-19: Seven scenarios. https://papers.ssrn.com/sol3/papers.cfm?
abstract_id=3547729.
Nicola, M., Alsafi, Z., Sohrabi, C., Kerwan, A., Al-Jabir, A., Iosifidis, C., Agha, M. and
Agha, R., 2020. The income implications of the coronavirus and COVID-19
pandemic: A review. International Journal of Surgery, Vol. 78, pp. 185–193.
Siche, R., 2020. What is the impact of COVID-19 disease on agriculture?. Scientia
Agropecuaria, 11(1), pp.3–6.
Sumner, A., Hoy, C., and Ortiz-Juarez, E. (2020). Estimates of the impact of COVID-19
on global poverty. UNU-WIDER Working Paper 2020/43.
Toreno, M., 2020. Prepare food systems for a long-haul fight against COVID-19.
Available at https://www.ifpri.org/blog/prepare-food-systems-long-haul-fight-
against-covid-19. Accessed Oct. 15, 2020.
UN-Habitat and World Food Program, 2020. Impact of COVID-19 on livelihoods,
food security & nutrition in East Africa: Urban focus. Available at
https://unhabitat.org/sites/default/files/2020/08/wfp-0000118161_1.pdf.
30